Normative
Law, contract, or formal standard in scope
Research method
Keep the authority behind every claim visible.
The canon separates requirements, findings, frameworks, guidance, methods, examples, and local choices before any principle can inform report work.
Evidence classesN E1 E2 F G P X H
Classification, not a universal ranking.
A standard can govern behavior without proving unique perceptual superiority. A study can support an effect without governing a business context.
Synthesized empirical
Replication, convergence, or strong synthesis
Primary empirical
A primary study; independent support not assumed
Framework
Academic model for design or evaluation
Guidance
Institutional interpretation, not a requirement
Practitioner
Professional method treated as a hypothesis
Exemplar
Craft or influence, not a general rule
House convention
Local operational choice without superiority claim
Principle lifecycle01 02 03 04
Register
Identify the primary or authoritative source, access, limits, and verification state.
Extract & triangulate
Record actual claims, population, task, outcome, contrary evidence, and replication.
Synthesize
Write the narrowest candidate with scope, confidence, boundary, and evaluation.
Audit & revise
Test report work and readers; accept, condition, reject universally, or retire.
Narrower claims beat clean slogans.
- Not every report must request an action.
- No chart type is best across all tasks.
- Answer-first is conditional, not a cognitive law.
- Animation, minimal ink, and fixed signal rules are not universal defaults.
Identity is not extraction.
Official-record checks the issuer’s record. Citation-checked confirms identity and reported contribution. Full-text-reviewed records relevant pages or rules.
No source is yet full-text-reviewed in the initial register, so initial principles remain candidates.
biai-os dogfood
Field evidence, never inherited truth.
Representative recurring reports from biai-os supply realistic tasks, practices, failures, and readers.
Audits and reader studies will measure lookup and comparison accuracy, signal detection, qualification recall, confidence calibration, decision identification, accessibility, effort, and longitudinal recognition. Existing chart bans, IBCS adaptations, XmR rules, confidence labels, and report anatomy face the same review as outside claims.